A method of registering point cloud with CAD model based on line contour matching

Registration of point cloud with its CAD model can be used to detect products. To register point cloud with its CAD model, feature characters are used. In this article, a registration method based on line contour matching is proposed. The registration process includes two parts. The one is to extract line contour points (LCPs) from point cloud. The other is to match LCPs of point cloud with LCPs of CAD model. To extract LCPs from point cloud, a slice set is constructed to extract section line data (SLD) containing LCPs from point cloud. Then LCPs are identified by judging the feature end of SLD and analyzing normal angle sequence (NAS) of SLD. LCPs of CAD model are non-uniformly distributed. So to make the matching process more efficient and robust, two groups of LCPs are uniformly sampled. Then best fitting method is used to match the two groups of sampled LCPs to obtain transformation parameters. Verification experiments are conducted on three line contours. And the calculated transformation parameters are compared with reference values. The results show that the deviations of rotation angle are within (-0.9°~0.3°). And deviations of translation elements are within ±0.4mm.